Neuromorphic Analog Processor for Always-on Smart Sensing Systems

Article By : Anne-Françoise Pelé

Aspinity has raised $5.3 million in Series A funding to ramp up the deployment of ultra-low power always-on sensing...

West Virginia University spinout Aspinity Inc. announced it has raised $5.3 million in Series A funding to accelerate the deployment of its reconfigurable analog modular processor (RAMP), an ultra-low power, analog processing chip designed to detect, analyze and classify raw analog sensor data. 10x less power Demand for always-on sensing devices is booming, with nearly one billion always-on voice first devices expected to enter the market by 2023, according to SAR Insight & Consulting. In anticipation, Aspinity has developed an “analyze-first-in-analog” approach which is claimed to reduce the power required at the edge by up to 10x and the volume of data handled by up to 100x for always-on applications.
Aspinity RAMP IC
Aspinity’s RAMP IC (Image: Aspinity)
Launched in June 2019, Aspinity’s RAMP IC analyzes the analog sensor data right as it enters the device and keeps higher-power downstream digital components in a deep sleep mode unless an event of interest is detected, Thomas Doyle, CEO and founder of Aspinity, explained to EE Times. “The RAMP IC is able to determine which data are important up front, before the data are digitized, so we completely eliminate the high-power digitization and analysis of irrelevant data, saving significant battery life.” Aspinity was founded in 2015, but the startup origin takes us back to 2012, when David Graham, a West Virginian University associate professor of computer science and electrical engineering, received the National Science Foundation Career Award grant for his research on energy-efficiency sensor networks using analog signal processing. As a university spin-off, Aspinity has exclusive and full rights to use the technology. $8.8 million This round of financing, led by Anzu Partners with participation of Amazon’s Alexa Fund, Birchmere Ventures, Mountain State Capital, and Riverfront Ventures, aims to strengthen Aspinity’s technical and partner support teams, to scale volume production to meet customer demand, and to accelerate the startup’s product roadmap. Asked to detail the startup’s plans in the near future, Doyle said Aspinity will primarily grow the team. “We’re building out both our engineering, operations, and applications teams in order to support customer design-ins, customize our RAMP [solution] for specific customers, and to support chip production, product qualification, etc.” Also on the roadmap is the start of volume production of Aspinity’s RAMP IC. “Our chip is manufactured on a mature technology node, so this helps with the efficiency of ramping production,” Doyle noted. “We are planning to be in production in the first half of 2021.” The Pittsburgh-based startup claims its solution is suitable for battery-operated, always-on sensing devices for consumer, smart home, IoT, and industrial markets. The RAMP’s analog blocks can be reprogrammed with application-specific algorithms to detect different events, different sensor inputs, and multiple types of output to support various applications, Doyle said. “Today, we have a software library that supports voice, alarm, glass break and other events. We will continue to build out this library to support many applications, but ultimately we will provide a programming platform so that customers can very easily write their own algorithms and ‘train’ the RAMP for their custom application.” Commenting on the Series A financing, Doyle specified, “Our total funding to date is $8.8 million, and this is actually Amazon’s third investment in Aspinity – the first was the selection of Aspinity for the inaugural Alexa accelerator and then through both our seed and Series A rounds of funding.” Asked more specifically about existing ties with Amazon beyond above-mentioned investments, Doyle said Aspinity demonstrated at CES 2020 that its voice activity detection solution could “detect speech and wake up hardware and the Amazon wake word engine (WWE) without reducing the accuracy of WWE.” He further commented: “Aspinity’s RAMP technology can significantly improve the battery life of many Alexa-enabled battery-operated devices which have gone beyond responding to voice but also to other sounds that RAMP can detect. All of these are driving Amazon’s interest in Aspinity.” ST, Infineon collaborations Lately, Aspinity has been collaborating with two major European semiconductor companies. In December 2019, Aspinity claimed it had demonstrated an ultra-low-power analog voice wake-up system for battery-operated devices with STMicroelectronics’ STM32H7 and STM32L4 ultra-high performance and power-efficient MCUs for connected and non-connected applications.
Thomas Doyle
Aspinity’s CEO Thomas Doyle
In May 2020, Aspinity and Infineon Technologies joined forces to accelerate the development of always-on sensing products with longer-lasting batteries. The combination of Aspinity’s technology and Infineon’s XENSIV sensors aims to facilitate a power-efficient analyze-first architecture in a new generation of power- and data-efficient always-on devices. “The combination of Infineon sensors and the Aspinity RAMP IC is targeted at small, battery-operated devices where always-on functionality cannot currently be integrated without having such a significant impact on battery life that the device is essentially not viable,” Doyle said. These two collaborations, he continued, “provide additional confidence to investors that the leading semiconductor manufacturers see significant value in Aspinity solutions. Most importantly, customers are interested in our collaborative solutions. So the funding that we received indirectly enables the relationships to move further in that we will be able to more quickly bring the RAMP chip into production to support these reference design customers for their upcoming products.”

Leave a comment